Subscriber access provided by Northern Illinois University
Article 3
Low-loss BaTiO-Si waveguides for nonlinear integrated photonics Felix Eltes, Daniele Caimi, Florian Fallegger, Marilyne Sousa, Eamon O’Connor, Marta D Rossell, Bert Offrein, Jean Fompeyrine, and Stefan Abel ACS Photonics, Just Accepted Manuscript • DOI: 10.1021/acsphotonics.6b00350 • Publication Date (Web): 22 Aug 2016 Downloaded from http://pubs.acs.org on August 27, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Photonics is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Photonics
Low-loss BaTiO3-Si waveguides for nonlinear integrated photonics Felix Eltes†*, Daniele Caimi†, Florian Fallegger†, Marilyne Sousa†, Eamon O’Connor†, Marta D. Rossell‡, Bert Offrein†, Jean Fompeyrine†, and Stefan Abel†* †
IBM Research – Zurich, Säumerstrasse 4, 8803 Rüschlikon, Switzerland Electron Microscopy Center, EMPA, Swiss Federal Laboratories for Materials Science and Technology, 8600 Dübendorf, Switzerland KEYWORDS. Barium titanate, silicon photonics, Pockels effect, propagation losses, hydrogenation. ‡
ABSTRACT: Barium titanate (BaTiO3) has become an attractive material to extend the functionalities of the silicon photonics platform because of its large Pockels coefficient of more than 1000 pm/V. BaTiO3 integrated epitaxially on silicon-on-insulator (SOI) substrates can be structured in passive, and electro-optic silicon photonic devices using slot-waveguide geometries, both of which have been demonstrated. However, all devices demonstrated so far suffer from high optical propagation losses of ~40 – 600 dB/cm, which limits their performance compared with state-of-the-art silicon photonics devices (< 2 dB/cm). Here, we identify the origin of these high propagation losses, and demonstrate a path to fabricate low-loss BaTiO3-Si waveguides with propagation losses of only 6 dB/cm. In particular, we identified the thin strontium titanate (SrTiO3) seed layer typically used for the epitaxial deposition of BaTiO3 on silicon as the main source of absorption: When manufacturing slot-waveguide structures, the BaTiO3/SrTiO3 layer stack is typically exposed to hydrogen, which is incorporated in the SrTiO3 layer, and causes absorption. We demonstrate that a low-temperature anneal is sufficient to remove hydrogen and to achieve low propagation losses in waveguides. Thus, we found a way to eliminate the previously observed showstopper for incorporating functional and highly nonlinear barium titanate films into silicon photonic structures, ultimately enabling ultra-high-speed switches and novel nonvolatile optical silicon photonic devices.
mum donor wafer size of 6 inch9,10, the integration of good electro-optical materials that are compatible with large-scale silicon wafers has recently attracted increased attention. One promising candidate is barium titanate, a ferroelectric oxide with a strong Pockels effect that can be deposited with high crystalline quality directly on silicon by, for example, molecular beam epitaxy (MBE).11 As recently demonstrated, BaTiO3 layers on silicon substrates retain a strong Pockels coefficient of r ≈ 150 pm/V12 and can be implemented in active and passive silicon photonic structures13–15. A major issue in all demonstrations of hybrid BaTiO3-Si devices so far has been the high propagation losses (αp) of 40 to > 500 dB/cm at the wavelength λ = 1.55 µm, much larger than the propagation losses below 2 dB/cm16 of state-of-the-art silicon waveguides. These high propagation losses were not expected because BaTiO3 bulk crystals are transparent at λ = 1.55 µm.17 In principle, these good optical properties are maintained in thin films, as shown by the rather low propagation losses of 4 dB/cm in waveguides made from BaTiO3 layers deposited on magnesium oxide substrates.18 Those results indicate that BaTiO3 thin films do not suffer intrinsically from high propagation losses, but that microstructural effects when grown on silicon, interface effects with silicon, or the processing conditions during waveguide fabrication introduce those high losses. In this work, we carefully studied various contributions to the propagation losses in hybrid BaTiO3-Si waveguides to ultimately identify and eliminate the reason for the losses. In particular, we used various waveguide geometries, process
Silicon photonics is becoming a mature platform for photonic integrated circuits, benefiting from the processing and fabrication knowledge of microelectronics to realize reliable nanoscale devices. To link the electric and optic domains, in general electro-optical modulators based on the plasmadispersion effect are used.1 Although high modulation speeds have been reached in Si modulators2, the devices typically suffer from relatively large insertion losses and the simultaneous change of the real and imaginary parts of the refractive index. Thus, advanced modulation formats, such as quadrature amplitude modulation (QAM), are difficult to implement. In telecommunication applications, these formats were enabled by exploiting the Pockels effect in LiNbO33,4, but silicon does not provide this option because of a vanishing Pockels effect5. Enabling the Pockels effect on the silicon photonic platform could lead to improved devices in terms of power consumption, insertion losses, and modulation speeds. In addition, it could allow functionalities that cannot be achieved with bulk silicon only, such as ultra-low-power tuning or nonvolatile optical memories. The technical options investigated include the application of strain6 and the use of silicon-organic hybrid structures7,8. These approaches suffer either from small Pockels coefficients or from low process compatibility due to thermal instability. Heterogeneous integration of inorganic materials is an alternative path to bring a strong Pockels effect to silicon photonics. However, as the integration of highquality LiNbO3 on silicon is limited to local bonding schemes with low yield or to wafer bonding approaches with a maxi1
ACS Paragon Plus Environment
ACS Photonics
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 13
Optical characterization We determined the propagation losses in the waveguide structures by measuring the transmission spectrum between 1.5 and 1.6 µm through waveguides with lengths varying from 100 µm to 9 mm. We used the combination of either a tunable laser with a power meter or a superluminescent light-emitting diode with a spectrum analyzer. To extract the propagation losses, αp, we first corrected for the spectral coupling characteristics of the grating couplers. Then, we linearly fitted the transmission versus waveguide length, using multiple waveguides for each length. As an example, Figure 5 shows transmission versus waveguide length, and the corresponding fit. The slope of the linear fit gives the propagation losses per unit length.24 The standard error was estimated based on the variance in losses of multiple (> 2) sets of waveguides with the same lengths, combined with the linear fit error. Chemical characterization The surface chemistry of BaTiO3 and SrTiO3 films was studied in an ultra-high-vacuum (UHV) XPS chamber with a base pressure of 10-9 mBar. The chamber is equipped with a monochromatic Al Kα X-ray source and a Phoibos150 hemispherical analyzer, mounted at 50° take-off angle, defined from the normal to sample surface. Gaussian–Lorentzian line shapes were used for deconvoluting the spectra after Shirley background subtraction.
conditions, and post-fabrication annealing treatments to separate loss channels arising from scattering and absorption in the different materials and at their interfaces. Typically, BaTiO3-Si waveguides are fabricated as halfway-etched slot-waveguide structures: The waveguide consists of a blanket BaTiO3 film sandwiched between the device Si of an SOI wafer and a patterned top Si layer (Figure 1a). The slot-waveguide geometry enhances the optical confinement in the BaTiO3 slab, while keeping bending losses low. A critical step in slot-waveguide fabrication is the integration of the top Si layer, which requires several deposition and etching steps. To disentangle the different origins of optical losses, we measured the propagation losses at various steps during the fabrication of the slot waveguides, using SiO2 strip-loaded waveguides (Figure 1b). Our geometry of choice exhibits a similar optical confinement in the BaTiO3 layer as that of the BaTiO3-Si waveguides (Figure 1). While the confinement factor is comparable, the bending losses in SiO2 strip-loaded waveguides are much higher, which prevents the geometry from being used for active photonic applications. However, the simple fabrication process makes these waveguides ideal for investigating propagation losses without the influence of processing.
METHODS Waveguide fabrication The BaTiO3 films used for waveguide fabrication were deposited by MBE on SOI substrates with 100 or 220 nm device silicon using the method reported in 13. A 4-nm-thin epitaxial SrTiO3 seed layer was grown via MBE prior to shuttered codeposition of BaTiO3, to ensure good crystalline quality and a low surface roughness of RRMS < 0.5 nm.12,19,20 Details of the deposition process are described elsewhere.13,21 We fabricated strip-loaded waveguides on these BaTiO3/SOI substrates by spin coating and patterning 170-nmthick hydrogen silsesquioxane (HSQ), which transforms into SiO2 when exposed via electron-beam lithography (EBL) (Figure 1). For the realization of the compact slot-waveguide structures, we bonded BaTiO3 films grown on SOI substrates with 220 nm device silicon onto SOI wafers with a 100-nm-thick device silicon layer via direct wafer bonding.22 We used 5-nmthick Al2O3 adhesion layers, deposited by atomic layer deposition (ALD) on both wafers prior to the bonding step.23 After bonding and annealing at 250°C, the initial MBE growth wafer was removed by grinding and selective wet etching, leaving only the device silicon as a top layer. This layer was thinned during the chemical etching process of the SOI wafer, leaving 205 nm Si. The silicon top layer was then structured using EBL and an HBr/O2 inductively coupled plasma-etching process to form waveguides (Figure 2).
RESULTS AND DISCUSSION To confirm the high propagation losses observed in BaTiO3Si slot-waveguide structures13–15, we reproduced these structures using the fabrication process described above. Indeed, the propagation losses of αp = 148±7 dB/cm for the TE mode are in agreement with the presence of an unknown loss channel in our structures. In principle, propagation losses can be caused by three different mechanisms: (1) Absorption, (2) scattering at waveguide sidewalls, interfaces and material defects, and (3) leakage. By careful simulations with wellestablished optical mode solvers (Phoenix and COMSOL Multiphysics), we were able to exclude radiation losses due to poor light confinement. To separate the contributions originating from absorption and scattering, we then analyzed αp for differently processed samples, ranging from unprocessed BaTiO3 films to fully processed slot waveguides, as described below. Process related losses SiO2 strip-loaded waveguides on BaTiO3/SOI substrates (Figure 3a) show the same propagation losses (2.6±0.2 dB/cm) as the reference waveguides fabricated without the BaTiO3 layer (2.9±0.2 dB/cm, Figure 3d). From this comparison, we can conclude that (1) MBE-grown BaTiO3 does not absorb at λ = 1.55 µm prior to any post-deposition etching, deposition or annealing steps; (2) light-scattering effects in the BaTiO3 layer can be disregarded – in agreement with the high structural quality described above (Figure 2), and (3) our experiments are limited by propagation losses of ~3 dB/cm due to scattering at the SiO2 sidewalls, assuming that in our reference waveguide there is only negligible absorption in the SiO2 and Si layers.25 The deposition of Al2O3 for wafer bonding increases the losses to 25.3±0.5 dB/cm on BaTiO3 substrates (Figure 3b), but does not alter the propagation losses of the SOI reference waveguides (2.8±0.3 dB/cm, Figure 3d). These results indicate that the Al2O3 layer itself is transparent, but that a chemical
Scanning Transmission Electron Microscopy (STEM) Bright-field STEM (BF-STEM) analysis, including cross sectioning and lamella preparation, were carried out by means of a FEI Helios Nanolab 450S focused ion beam (FIB) operated at an accelerating voltage 30 kV for BF-STEM. High-angle annular dark field STEM (HAADF-STEM) measurements were performed using a double spherical aberration-corrected JEOL JEM-ARM200F microscope operated at 200 kV. The convergence semiangle was set to 25.3 mrad, and the annular semi-detection range of the annular dark-field detector was set to collect electrons scattered between 90 and 370 mrad. 2
ACS Paragon Plus Environment
Page 3 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Photonics though oxygen vacancies could not have been cured owing to the low oxygen partial pressure (< 10-9 mBar). The losses obtained after the UHV anneal can be explained by the presence of two competing processes. First, outgassing of hydrogen strongly reduces the losses induced by the hydrogen plasma exposure. Second, the high temperature and low oxygen partial pressure cause absorbing oxygen vacancies, leading to moderate losses after UHV annealing. In a control experiment, another SiO2 strip-loaded BaTiO3/SOI sample without prior hydrogen plasma exposure (αp ≈ 3 dB/cm) indeed exhibited propagation losses of 90±11 dB/cm after an UHV anneal at a higher temperature (600°C instead of 550°C). The results of the UHV annealing experiments reveal two important aspects: First, long anneals at low oxygen partial pressure induce propagation losses due to the formation of oxygen vacancies. Second, high propagation losses induced by hydrogen exposure can be strongly reduced through UHV annealing. We assume that the hydrogen outgasses from the BaTiO3/SrTiO3 layer stack, resulting in reduced propagation losses. To avoid the formation of oxygen vacancies while outgassing hydrogen, we finally annealed SiO2 strip-loaded BaTiO3Si waveguides with high αp (> 300 dB/cm), from hydrogen plasma exposure, in oxygen at 500°C for 60 min, which resulted in waveguides with αp = 2.2±0.2 dB/cm. As hightemperature oxygen annealing is not compatible with many complementary metal-oxide-semiconductor (CMOS) fabrication steps, we adjusted the annealing process to more moderate conditions. Ultimately, we were able to successfully outgas hydrogen from high-loss waveguides by annealing in argon (pO2 < 10-7 mBar) at 350°C, which is compatible with standard CMOS processes. At this annealing temperature, no oxygen vacancies are induced, resulting in 2.6±0.2 dB/cm propagation losses.
reaction with the BaTiO3 or SrTiO3 layer occurs during the ALD process. One possible reactant is hydrogen, which is formed during the ALD process from H2O precursors26. The strongly reducing hydrogen might create oxygen vacancies in SrTiO3 or BaTiO3, resulting in higher propagation losses.27 After fabrication of the BaTiO3-Si slot-waveguide layer stack by bonding a crystalline silicon top layer (Figure 3c), the propagation losses increase to 154±26 dB/cm, which is in the same range as those of the fully processed waveguides (Figure 1a). Here again, we attribute the origin of this strong increase to hydrogen: During the bonding process, excess water is formed, which decomposes into hydrogen when oxidizing the Si below the Al2O3 layer.22,28 As described above, the hydrogen might react with BaTiO3 or SrTiO3 – in particular because of the increased diffusion into the BaTiO3/SrTiO3 stack at the elevated temperatures during the annealing process. Reference waveguides with nonabsorbing Si3N4 layers25 instead of the BaTiO3 film (Figure 3f) show significantly lower propagation losses (18±2 dB/cm). The refractive index of the Si3N4 was tuned to a similar value as that of the BaTiO3 layer to ensure the same mode profile. The lower losses in the Si3N4 slot waveguides indicate that possible scattering losses at the bonding interface contribute only slightly to the overall much higher propagation losses of BaTiO3 slot waveguides. The comparison of different waveguides and process steps provides strong indications that hydrogen induces absorbing defects in the BaTiO3/SrTiO3 layer stack during the bonding process. Absorption caused by hydrogen exposure is also consistent with the previously observed high losses in BaTiO3-Si slot-waveguide structures fabricated from amorphous silicon (a-Si) top layers deposited via plasma-enhanced chemical vapor deposition (PECVD)15: During the deposition of lowloss a-Si, SiH4 precursors and excess H2 are used29,30, which produces reactive atomic hydrogen during deposition. The dependence of the hydrogen concentration on the specific process conditions used might also explain the large variations in propagation losses observed in the past (40 – 600 dB/cm13–15).
Chemical sensitivity to hydrogen To understand the physical process behind the hydrogeninduced absorption better, we correlated the observed losses with chemical information from the uppermost 5 – 10 nm of a SrTiO3 and a BaTiO3-terminated sample via XPS measurements. After exposing the sample to a hydrogen plasma, a much stronger chemical change was observed in the SrTiO3 layer than in the BaTiO3 layer. In particular, the Ti 2p peak reveals a significant reduction of Ti4+ to Ti3+ in SrTiO3, whereas in BaTiO3 the fraction of Ti3+ remains unchanged (Figure 4). These results indicate that the hydrogen selectively reduces the titanium in the SrTiO3 layer, which leads to an increase in the optical absorption.
Role of Hydrogenation To confirm the effect of hydrogen exposure, we performed control experiments by explicitly exposing SiO2 strip-loaded BaTiO3/SOI waveguides to hydrogen plasma. The plasma exposure was performed at 290°C under 80 mTorr pressure and 100 sccm flow of hydrogen with 200 W plasma power in an Oxford FlexAL ALD tool. Identical conditions were used for all hydrogen plasma exposures in this study. Indeed, hydrogen exposure induces propagation losses in the waveguide structures that increase nonlinearly with the process time up to 320±20 dB/cm (Table 1). The exposure of a SOI reference sample with SiO2 waveguides to the same conditions showed that the plasma has no influence on the SiO2 and silicon layers, and confirms that the higher losses are due to the presence of the BaTiO3/SrTiO3 layer stack in the waveguide (Table 1). While the results of the plasma exposure experiments confirm that exposing BaTiO3/SrTiO3 to atomic hydrogen induces absorption, they do not reveal whether the absorption is due to the creation of oxygen vacancies or due to another interaction with hydrogen, such as hydrogenation of the materials31–33. To investigate the role of oxygen vacancies, we first prepared SiO2 strip-loaded BaTiO3/SOI waveguides with high propagation losses (> 300 dB/cm) using a long hydrogen plasma exposure (same conditions as above), and subsequently annealed the sample in UHV (< 10-9 mBar) at ~550°C for 60 min. This UHV annealing step strongly reduced αp to 49±5 dB/cm, even 3
ACS Paragon Plus Environment
ACS Photonics
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 13
Table 1. Influence of hydrogen plasma exposure on propagation losses in SiO2-strip-loaded BaTiO3 and Si waveguides. Waveguide layers
Hydrogen plasma time
Propagation losses [dB/cm]
50 nm BaTiO3 / 4 nm SrTiO3/ 100 nm Si / 2 µm SiO2
None
2.6±0.2
50 nm BaTiO3 / 4 nm SrTiO3/ 100 nm Si / 2 µm SiO2
40 min
50±3
50 nm BaTiO3 / 4 nm SrTiO3/ 100 nm Si / 2 µm SiO2
80 min
320±20
220 nm Si / 2 µm SiO2
None
1.8±0.3
220 nm Si / 2 µm SiO2
80 min
1.7±0.1
To verify that the 4-nm-thick SrTiO3 layer is the main source of absorption in our structures, we compared SiO2 waveguides made using a 50 nm BaTiO3/4 nm SrTiO3 layer stack with waveguides using only a 4 nm SrTiO3 layer. Both stacks where fully oxygenated prior to waveguide fabrication. Assuming that both BaTiO3 and SrTiO3 contribute significantly to the absorption after hydrogenation, one would expect much higher propagation losses on the BaTiO3/SrTiO3 sample because of the high optical confinement factor of 13.2% in the BaTiO3/SrTiO3 layer stack. The much lower optical confinement in the SrTiO3 layer (1.4%) of our SrTiO3-layer-only reference sample would accordingly result in significantly lower losses. However, after a 20-min hydrogen plasma treatment, the propagation losses in both samples increase similarly from 2.6±0.2 dB/cm to 50±3 dB/cm for the BaTiO3/SrTiO3 sample (Table 1), and from 4.0±0.5 dB/cm to 81±3 dB/cm in the SrTiO3-layer-only sample. These results show that the main contribution of the propagation losses indeed originates from the 4-nm-thick SrTiO3 layer rather than from the BaTiO3 layer, and confirm the selectivity of the hydrogenation process observed with XPS. Taking into account the optical confinement factor and the overall waveguide losses, we can estimate the losses in the SrTiO3 layer as ~5,000 dB/cm after 20-min hydrogen plasma exposure. Assuming that the high propagation losses in all BaTiO3/SrTiO3/Si waveguides are dominated by absorption in SrTiO3, we consistently estimate that the losses in the SrTiO3 layer induced during the slot-waveguide fabrication are more than 10,000 dB/cm.
layer as the major source of these large propagation losses. The absorption is induced by hydrogen exposure during the integration of the top silicon layer, which is done by either direct wafer bonding or PECVD. In contrast, absorption effects in the BaTiO3 layer are negligible. We developed CMOS-compatible annealing procedures that fully recover the hydrogen-induced losses and that can be used to obtain lowloss slot-waveguide structures (αp = 6 dB/cm, Figure 7) and high-Q ring resonator cavities (Q > 20,000, Figure 6). The ability to remove high propagation losses puts renewed focus on a BaTiO3-based photonic technology, and renders it very appealing for more advanced silicon photonic structures: It enables novel applications and structures, such as high-speed modulators, non–volatile optical memories, and radiofrequency electric field detectors.
AUTHOR INFORMATION Corresponding Authors * E-mail:
[email protected] * E-mail:
[email protected] Notes The authors declare no competing financial interest.
ACKNOWLEDGMENTS Financial support from the European Commission under project FP7-ICT-2013-11-619456 SITOGA, as well as from the Swiss National Foundation under project 200021_159565 PADOMO is acknowledged.
Fabrication of low loss slot-waveguide structures Finally, we applied our insights into the loss mechanism and the curing procedure to fabricate low-loss BaTiO3-based slotwaveguide structures. Using the same fabrication scheme as described above, we were able to reduce the propagation losses from initially αp = 148±7 dB/cm to 6.3±0.9 dB/cm by annealing the waveguides in oxygen at 350°C for 60 min (Figure 5). The remaining propagation losses are within the typical range of slot-waveguide structures, which generally suffer from scattering effects at the interfaces.36,37 Using this fabrication method, it is now possible to fabricate various optical elements, such as couplers, splitters and interferometers, with a highly nonlinear optical crystal embedded, in silicon photonics. Ring resonators with a radius of 75 µm with well-defined resonances and a quality factor of Q > 20,000 (Figure 6) demonstrate the usability of the BaTiO3-Si hybrid waveguides.
REFERENCES (1) Reed, G. T.; Mashanovich, G.; Gardes, F. Y.; Thomson, D. J. Silicon Optical Modulators. Nat. Photonics 2010, 4, 518–526. (2) Xu, H.; Li, X.; Xiao, X.; Zhou, P.; Li, Z.; Yu, J.; Yu, Y. HighSpeed Silicon Modulator with Band Equalization. Opt. Lett. 2014, 39, 4839–4842. (3) Mahapatra, A.; Murphy, E. J. Electrooptic Modulators; Kaminow, I. P., Tingye, L., Eds.; Elsevier, 2002. (4) Wooten, E. L.; Kissa, K. M.; Yi-Yan, A.; Murphy, E. J.; Lafaw, D. A.; Hallemeier, P. F.; Maack, D.; Attanasio, D. V.; Fritz, D. J.; McBrien, G. J.; Bossi, D. E. A Review of Lithium Niobate Modulators for Fiber-Optic Communications Systems. IEEE J. Sel. Top. Quantum Electron. 2000, 6, 69–82. (5) Soref, R.; Bennett, B. Electrooptical Effects in Silicon. IEEE J. Quantum Electron. 1987, 23, 123–129. (6) Chmielak, B.; Waldow, M.; Matheisen, C.; Ripperda, C.; Bolten, J.; Wahlbrink, T.; Nagel, M.; Merget, F.; Kurz, H. Pockels Effect Based Fully Integrated, Strained Silicon Electro-Optic Modulator. Opt. Express 2011, 19, 17212. (7) Alloatti, L.; Korn, D.; Palmer, R.; Hillerkuss, D.; Li, J.; Barklund, A.; Dinu, R.; Wieland, J.; Fournier, M.; Fedeli, J.; Yu, H.; Bogaerts, W.; Dumon, P.; Baets, R.; Koos, C.; Freude, W.; Leuthold, J. 42.7 Gbit/s Electro-Optic Modulator in Silicon Technology. Opt. Express 2011, 19, 11841–11851.
CONCLUSIONS Previous reports on novel BaTiO3-Si hybrid waveguides consistently showed high propagation losses of up to 600 dB/cm. In this study, we identified a strong optical absorption of more than 10,000 dB/cm in the thin SrTiO3 seed 4
ACS Paragon Plus Environment
Page 5 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Photonics
(8) Ding, R.; Baehr-Jones, T.; Liu, Y.; Bojko, R.; Witzens, J.; Huang, S.; Luo, J.; Benight, S.; Sullivan, P.; Fedeli, J. M.; Fournier, M.; Dalton, L.; Jen, A.; Hochberg, M. A Low VπL Modulator with GHz Bandwidth Based on an Electro-Optic Polymer-Clad Silicon Slot Waveguide. Proc. 7th IEEE International Conference on Group IV Photonics, pp 201–203 (IEEE 2010). (9) Chen, L.; Chen, J.; Nagy, J.; Reano, R. M. Highly Linear Ring Modulator from Hybrid Silicon and Lithium Niobate. Opt. Express 2015, 23, 13255. (10) Rabiei, P.; Ma, J.; Khan, S.; Chiles, J.; Fathpour, S. Heterogeneous Lithium Niobate Photonics on Silicon Substrates. Opt. Express 2013, 21, 114–115. (11) McKee, R. a.; Walker, F. J.; Conner, J. R.; Specht, E. D.; Zelmon, D. E. Molecular Beam Epitaxy Growth of Epitaxial Barium Silicide, Barium Oxide, and Barium Titanate on Silicon. Appl. Phys. Lett. 1991, 59, 782–784. (12) Abel, S.; Stöferle, T.; Marchiori, C.; Rossel, C.; Rossell, M. D.; Erni, R.; Caimi, D.; Sousa, M.; Chelnokov, A.; Offrein, B. J.; Fompeyrine, J. A Strong Electro-Optically Active Lead-Free Ferroelectric Integrated on Silicon. Nat. Commun. 2013, 4, 1671. (13) Abel, S. Electro-Optic Photonic Devices Based on Epitaxial Barium Titanate Thin Films on Silicon, Ph.D. dissertation, Univ. Grenoble, France, 2014. (14) Abel, S.; Stoeferle, T.; Marchiori, C.; Caimi, D.; Czornomaz, L.; Stuckelberger, M.; Sousa, M.; Offrein, B.; Fompeyrine, J. A Hybrid Barium Titanate-Silicon Photonics Platform for UltraEfficient Electro-Optic Tuning. J. Light. Technol. 2015, 34, 1688– 1693. (15) Xiong, C.; Pernice, W.; Ngai, J.; Reiner, J.; Kumarh, D.; Walker, F.; Ahn, C.; Tang, H. Active Silicon Integrated Nanophotonics : Ferroelectric BaTiO3 Devices. Nano Lett. 2014, 14, 1419–1425. (16) Bogaerts, W.; Selvaraja, S. K. Compact Single-Mode Silicon Hybrid Rib/strip Waveguide with Adiabatic Bends. IEEE Photonics J. 2011, 3, 422–432. (17) Cardona, M. Optical Properties and Band Structure of SrTiO3 and BaTiO3. Phys. Rev. 1965, 140 (2A), 651–655. (18) Gill, D. M.; Block, B. A.; Conrad, C. W.; Wessels, B. W.; Ho, S. T. Thin Film Channel Waveguides Fabricated in Metalorganic Chemical Vapor Deposition Grown BaTiO3 on MgO. Appl. Phys. Lett. 1996, 69, 2968–2970. (19) Yu, Z.; Ramdani, J.; Curless, J. A.; Overgaard, C. D.; Finder, J. M.; Droopad, R.; Eisenbeiser, K. W.; Hallmark, J. A.; Ooms, W. J.; Kaushik, V. S. Epitaxial Oxide Thin Films on Si(001). J. Vac. Sci. Technol. B Microelectron. Nanom. Struct. 2000, 18, 2139–2145. (20) Niu, G.; Yin, S.; Saint-Girons, G.; Gautier, B.; Lecoeur, P.; Pillard, V.; Hollinger, G.; Vilquin, B. Epitaxy of BaTiO3 Thin Film on Si(001) Using a SrTiO3 Buffer Layer for Non-Volatile Memory Application. Microelectron. Eng. 2011, 88, 1232–1235. (21) Abel, S.; Sousa, M.; Rossel, C.; Caimi, D.; Rossell, M. D.; Erni, R.; Fompeyrine, J.; Marchiori, C. Controlling Tetragonality and Crystalline Orientation in BaTiO3 Nano-Layers Grown on Si. Nanotechnology 2013, 24, 285701. (22) Tong, Q. Y.; Cha, G.; Gafiteanu, R.; Gosele, U. Low Temperature Wafer Direct Bonding. Microelectron. Reliab. 1994, 3, 29–35. (23) Daix, N.; Uccelli, E.; Czornomaz, L.; Caimi, D.; Rossel, C.; Sousa, M.; Siegwart, H.; Marchiori, C.; Hartmann, J. M.; Shiu, K.-T.; Cheng, C.-W.; Krishnan, M.; Lofaro, M.; Kobayashi, M.; Sadana, D.; Fompeyrine, J. Towards Large Size Substrates for III-V CoIntegration Made by Direct Wafer Bonding on Si. APL Mater. 2014, 2, 086104. (24) Vlasov, Y.; McNab, S. Losses in Single-Mode Silicon-onInsulator Strip Waveguides and Bends. Opt. Express 2004, 12, 1622– 1631. (25) Handbook of Optical Constants of Solids; Palik, E. D., Ed.; Academic Press: Burlington, 1997. (26) Ylivaara, O. M. E.; Liu, X.; Kilpi, L.; Lyytinen, J.; Schneider, D.; Laitinen, M.; Julin, J.; Ali, S.; Sintonen, S.; Berdova, M.; Haimi, E.; Sajavaara, T.; Ronkainen, H.; Lipsanen, H.; Koskinen, J.; Hannula, S. P.; Puurunen, R. L. Aluminum Oxide from Trimethylaluminum and Water by Atomic Layer Deposition: The
Temperature Dependence of Residual Stress, Elastic Modulus, Hardness and Adhesion. Thin Solid Films 2014, 552, 124–135. (27) Jalan, B.; Engel-Herbert, R.; Mates, T. E.; Stemmer, S. Effects of Hydrogen Anneals on Oxygen Deficient SrTiO3-X Single Crystals. Appl. Phys. Lett. 2008, 93, 052907. (28) Gösele, U.; Tong, Q. Y.; Schumacher, A.; Kräuter, G.; Reiche, M.; Plößl, A.; Kopperschmidt, P.; Lee, T. H.; Kim, W. J. Wafer Bonding for Microsystems Technologies. Sensors Actuators, A Phys. 1999, 74, 161–168. (29) de Dood, M. J. A.; Polman, A.; Zijlstra, T.; Van Der Drift, E. W. J. M. Amorphous Silicon Waveguides for Microphotonics. J. Appl. Phys. 2002, 92, 649–653. (30) Selvaraja, S. K.; Sleeckx, E.; Schaekers, M.; Bogaerts, W.; Thourhout, D. Van; Dumon, P.; Baets, R. Low-Loss Amorphous Silicon-on-Insulator Technology for Photonic Integrated Circuitry. Opt. Commun. 2009, 282, 1767–1770. (31) Ito, T. U.; Higemoto, W.; Matsuda, T. D.; Koda, A.; Shimomura, K. Shallow Donor Level Associated with Hydrogen Impurities in Undoped BaTiO3. Appl. Phys. Lett. 2013, 103, 2011– 2015. (32) Varley, J. B.; Janotti, A.; Van De Walle, C. G. Hydrogenated Vacancies and Hidden Hydrogen in SrTiO3. Phys. Rev. B - Condens. Matter Mater. Phys. 2014, 89, 1–8. (33) Yukawa, H.; Nakatsuka, K.; Morinaga, M. Electronic Structures of Hydrogen in Perovskite-Type Oxide, SrTiO3. Solid State Ionics 1999, 116, 89–98. (34) Lin, C.; Posadas, A.; Hadamek, T.; Demkov, A. A. Final-State Effect on X-Ray Photoelectron Spectrum of Nominally d1 and nDoped d0 Transition-Metal Oxides. Phys. Rev. B 2015, 92, 035110. (35) Wang, J. L.; Leroy, J.; Niu, G.; Saint-Girons, G.; Gautier, B.; Vilquin, B.; Barrett, N. Chemistry and Structure of BaTiO3 UltraThin Films Grown by Different O2 Plasma Power. Chem. Phys. Lett. 2014, 592, 206–210. (36) Baehr-Jones, T.; Hochberg, M.; Walker, C.; Scherer, A. HighQ Optical Resonators in Silicon-on-Insulator-Based Slot Waveguides. Appl. Phys. Lett. 2005, 86, 081101. (37) Sun, R.; Dong, P.; Feng, N.; Hong, C.; Michel, J.; Lipson, M.; Kimerling, L. Horizontal Single and Multiple Slot Waveguides: Optical Transmission at Lambda = 1550 nm. Opt. Express 2007, 15, 17967–17972. (38) Castera, P.; Tulli, D.; Gutierrez, A. M.; Sanchis, P. Influence of BaTiO3 Ferroelectric Orientation for Electro-Optic Modulation on Silicon. Opt. Express 2015, 23, 15332.
ACS Paragon Plus Environment
5
1 2 3 4 5 6
(b)
220ACS nm Si Photonics 170 nm SiO2 50 nm BaTiO3 4 nm SrTiO3 100 nm Si
Page 6 of 13
2 µm SiO2
ACS Paragon Plus Environment |Sz|
(a)
ΓBaTiO3 = 14%
ΓBaTiO3 = 13%
(c)
Page 7 of 13 (a)
ACS Photonics
Si SiO2
1 SiO2 Pt 4 nm SrTiO3 2 BaTiO3 3 205 nm Si 4 (b) 5 50 nm BaTiO3 6 100 nm Si 10 nm Al2O3 7 200 nm 8 2 µm SiO2 ACS Paragon Plus Environment SiO2 9 10 Si
5 nm BaTiO3
10 nm
(a)
(b)
BaTiO3 waveguides
ACS Photonics
Reference waveguides
1 2 3 4 5 6 7 8 9 10 11 12 13
2.6 dB/cm
(c)
Page 8 of 13
25 dB/cm
150 dB/cm
Slot waveguide process flow
(d)
(e)
(f)
ACS Paragon Plus Environment 2.9 dB/cm
2.8 dB/cm
18 dB/cm
As grown
Page 9 of 13
After Hydrogen
ACS Photonics Measured Fit
SrTiO3
Ti4+ Ti3+
BaTiO3
Intensity [a. u.]
1 2 3 4 5 6 7 8 9 10 11 12
ACS Paragon Plus Environment 460
458
456
460
Binding Energy [eV]
458
456
Transmitted power [dBm]
0
ACS Photonics of 13 Not annealed,Page αp = 14810 dB/cm Annealed, αp= 6 dB/cm Linear fit
-10
1 2 -20 3 4 -30 5 6 -40 7 8 -50 9 10 11 0 12
220 nm Si 50 nm BaTiO3 4 nm SrTiO3 100 nm Si 2 µm SiO ACS Paragon Plus Environment
1
2
3
4
Waveguide length [mm]
5
2
6
Page 11 of 13
ACS Photonics Not annealed, αp= 148 dB/cm Annealed, αp= 6 dB/cm
0
Transmission [dB]
1 2 -5 3 4 -10 5 6 -15 7 8 -20 9 10 11 -2 12
ACS Paragon Plus Environment -1
0
Δλ [nm]
1
2
Propagation losses [dB/cm]
600 400
ACS Photonics
Page 12BaTiO of 13 3
MgO
BaTiO3 on MgO Si
BaTiO 1 SiO 2 200 BaTiO3 on SOI 3 4 50 5 6 25 7 8 9 0 ACS Paragon Plus Environment 10 5 4 14 96 ork 01 01 9 20 11 Gill 1 l2 g2 is w e el n h b b T A A Xio 12 3
2
H2
H2
Page 13 of 13
3 ACS BaTiO Photonics Si
1 2 3 4
Transmission
SiO2
148 dB/cm
6 dB/cm
ACS Paragon Plus Environment Wavelength
Wavelength
